GenAI in General
An e-commerce company is designing a chatbot to assist customers with product inquiries, order status updates, and troubleshooting. Which capability of AI is most relevant for ensuring that the chatbot provides real-time responses to customer queries?
a. Adaptability
b. Responsiveness
c. Simplicity
d. Data Efficiency
Correct Answer: B
Generative AI models can produce content in real-time, leading to quick response times and dynamic interactions. This is especially helpful for chatbots, virtual assistants, and other interactive applications that necessitate immediate responses. Responsiveness ensures that the chatbot generates content in real-time, providing immediate answers to customer queries. In an e-commerce context, customers expect quick assistance, making responsiveness critical.
Which statement best describes the concept of an AWS region?
a. An AWS Region represents the country where the AWS infrastructure exist
b. An AWS Region is a geographical location with a collection of Availability Zones
c. An AWS Region is a geographical location with a collection of Edge locations
d. An AWS Region is a virtual network dedicated only to a single AWS customer
Correct Answer: B
An AWS Region is a physical location in the world. Each region has multiple, isolated locations known as Availability Zones. Availability Zones consist of one or more discrete data centers, each with redundant power, networking, and connectivity. These Availability Zones offer you the ability to operate production applications and databases that are more highly available, fault tolerant, and scalable than would be possible to operate out of a single data center. Also, each AWS Region is designed to be completely isolated from the other AWS Regions. This achieves the greatest possible fault tolerance and stability.
Which of the following does NOT belong to the AWS Cloud Computing models?
a. Infrastructure as a Service (IaaS)
b. Software as a Service (SaaS)
c. Platform as a Service (PaaS)
d. Networking as a Service (NaaS)
Correct Answer: D
There are three Cloud Computing Models, networking is part of IaaS.
A financial company is using an AI model to identify potential loan defaults. To ensure the model works well in production, they must set up processes for capturing real-time data, comparing it with the training set, detecting performance issues, and generating alerts. Which stage of the model development pipeline should the company focus on?
a. Model Training
b. Model Evaluation
c. Model Monitoring
d. Data Collection
Correct Answer: C
The model monitoring system must capture data, compare that data to the training set, define rules to detect issues and send alerts. This process repeats on a defined schedule when initiated by an event or when initiated by human intervention. The issues detected in the monitoring phase include data quality, model quality, bias drift, and feature attribution drift.
A company uses a generative AI model to automate content creation for its social media posts. The model has been trained on various topics, including technology, fashion, and travel. Recently, the company decided to expand its business into the fitness industry. Which capability of generative AI will be crucial for the model to adjust its output and generate relevant fitness-related content?
a. Scalability
b. Adaptability
c. Responsiveness
d. Creativity and exploration
Correct Answer: B
Generative AI models need adaptability to adjust their output when faced with new domains. By learning from data and generating context-specific content, they can tailor their responses to relevant topics. In this scenario, the model’s ability to adapt ensures it can create fitness-related social media content effectively.
Why does every AWS Region contain multiple Availability Zones?
a. Multiple Availability Zones allows for data replication and global reach
b. Multiple Availability Zones results in lower total cost compared to deploying in a single Availability Zone
c. Multiple Availability Zones within a region increases the storage capacity available in that region
d. Multiple Availability Zones allows you to build resilient and highly available architectures
Correct Answer: D
Resilience is the ability of an architecture to continue providing the same quality of service even if some of its resources become inaccessible. Deploying your resources across multiple Availability Zones offer you the ability to operate production applications and databases that are more resilient, highly available, and scalable than would be possible from a single data center.
What are the benefits of having infrastructure hosted in AWS? (Choose TWO)
a. Operating applications on behalf of customers
b. There is no need to worry about security
c. Increasing speed and agility
d. Gaining complete control over the physical infrastructure
e. All of the physical security and most of the data/network security are taken care of for you
Correct Answer: C, E
All of the physical security are taken care of for you. Amazon data centers are surrounded by three physical layers of security. “Nothing can go in or out without setting off an alarm”. It’s important to keep bad guys out, but equally important to keep the data in which is why Amazon monitors incoming gear, tracking every disk that enters the facility. And “if it breaks we don’t return the disk for warranty. The only way a disk leaves our data center is when it’s confetti.”
Most (not all) data and network security are taken care of for you. When we talk about the data/network security, AWS has a “shared responsibility model” where AWS and the customer share the responsibility of securing them. For example, the customer is responsible for creating rules to secure their network traffic using the security groups and is also responsible for protecting data with encryption.
"Increasing speed and agility" is also a correct answer because in a cloud computing environment, new IT resources are only a click away, which means it requires less time to make those resources available to developers - from weeks to just minutes. This results in a dramatic increase in agility for the organization, since the cost and time it takes to experiment and develop is significantly lower.
A development team is building a social media platform that allows users to upload images. They want to ensure that inappropriate content is filtered out. Which Amazon Rekognition API can detect unsafe content in a specified JPEG or PNG format image?
a. DetectLabels
b. DetectModerationLabels
c. DetectFaces
d. DetectText
Correct Answer: B
The DetectModerationLabels API is specifically designed for content moderation. It analyzes images to identify unsafe or inappropriate content, such as nudity, violence, or suggestive material. This API’s primary purpose is to detect moderation labels, making it suitable for filtering out unsafe content. When building a social media platform or any application that allows users to upload images, you can use DetectModerationLabels to flag or block content that violates community guidelines automatically. It provides a list of moderation labels associated with the image, along with confidence scores.
An organization plans to implement an artificial intelligence (AI) system to assess and recommend individuals for eligibility in various public health initiatives and social welfare programs. The system analyzes data from multiple sources, including census data, employment records, and financial information. The organization needs to streamline the application process and ensure that eligible individuals receive the support they need.
Which core dimension of responsible AI should the organization prioritize to ensure that the machine learning model aligns with ethical principles and provides clarity on how decisions are made?
a. Transparency
b. Fairness
c. Safety
d. Privacy and Security
Correct Answer: A
Responsible AI refers to the practice of developing and deploying AI systems in a way that promotes ethical principles, such as fairness, transparency, privacy, and accountability. It aims to mitigate potential risks and negative impacts associated with AI systems. Transparency refers to the ability to understand and inspect the inner workings, decision-making processes, and outputs of an AI system. It involves making the AI system’s behavior, decisions, and underlying logic visible and comprehensible to relevant stakeholders, such as developers, regulators, and end-users.
You want to create a backup of your data in another geographical location. Where should you create this backup?
a. In another Local Zone
b. In another Region
c. In another Availability Zone
d. In another Edge location
Correct Answer: B
A Region is a physical location around the world where AWS clusters data centers. AWS calls each group of logical data centers an Availability Zone. Each AWS Region consists of multiple, isolated, and physically separate Availability Zones within a geographic area. Unlike other cloud providers, who often define a region as a single data center, the multiple Availability Zones design of every AWS Region offers advantages for customers. Each Availability Zone has independent power, cooling, and physical security and is connected via redundant, ultra-low-latency networks. AWS customers focused on high availability can design their applications to run in multiple Availability Zones to achieve even greater fault-tolerance.
Your application has recently experienced significant global growth, and international users are complaining of high latency. What is the AWS characteristic that can help improve your international users’ experience?
a. Elasticity
b. Data durability
c. Global reach
d. High availability
Correct Answer: C
With AWS, you can deploy your application in multiple regions around the world. The user will be redirected to the Region that provides the lowest possible latency and the highest performance. You can also use the CloudFront service that uses edge locations (which are located in most of the major cities across the world) to deliver content with low latency and high performance to your global users.
A company is considering implementing generative AI models to enhance its customer support system. However, the technical team is concerned about the potential disadvantages and limitations of using such models in production. They need to identify the specific challenges that may arise when working with generative AI. What are some common challenges or drawbacks associated with using generative AI models? (Select TWO.)
a. Hallucination
b. Knowledge Cutoff
c. Fraud detection
d. Low Recall
e. Personalization
Correct Answer: A, B
Generative AI models offer powerful capabilities for enhancing customer support systems by generating responses, handling inquiries, and providing automated assistance. However, deploying these models in a live production environment, particularly in customer support, requires careful consideration of potential challenges. Understanding the limitations of generative AI is essential to ensuring that the system can effectively and accurately serve customers.
A retail company wants to improve its customer engagement by providing personalized recommendations to online shoppers. They decide to leverage generative AI to achieve this goal.
Which characteristics of a Generative AI system are relevant for creating personalized product recommendations? (Select THREE.)
Correct Answer: B, D, F
Personalization is a critical capability for generative AI. It involves tailoring content or recommendations to individual users based on their preferences or characteristics, which enhances user experiences and engagement. Scalability can handle increasing workloads efficiently. It’s crucial for large-scale applications. Data efficiency relates to how well a model can learn from limited data.
A company has business critical workloads hosted on AWS and they are unwilling to accept any downtime. Which of the following is a recommended best practice to protect their workloads in the event of an unexpected natural disaster?
a. Deploy AWS resources across multiple Availability Zones within the same AWS Region
b. Replicate data across multiple Edge Locations worldwide and use Amazon CloudFront to perform automatic failover in the event of an outage
c. Create point-in-time backups in another subnet and recover this data when a disaster occurs
d. Deploy AWS resources to another AWS Region and implement an Active-Active disaster recovery strategy
Correct Answer: D
Disaster recovery is about preparing for and recovering from events that have a negative impact on your business continuity or finances. This could be a natural disaster, hardware or software failure, a network outage, a power outage, physical damage to a building like fire or flooding, or some other significant disaster. In AWS, customers have the flexibility to choose the disaster recovery approach that fits their budget. The approaches could be as minimum as backup and restore from another AWS Region or full-scale multi-region Active-Active solution. With the multi-region Active-Active solution, your workload is deployed to, and actively serving traffic from, multiple AWS Regions. If an entire Region goes down because of a natural disaster or any other reason, the other Regions will still be available and able to serve user requests.
Under the shared responsibility model, which of the following is the responsibility of AWS?
a. Configuring infrastructure devices
b. Client-side encryption
c. Filtering traffic with Security Groups
d. Server-side encryption
Correct Answer: A
Under the shared responsibility model, AWS is responsible for the hardware and software that run AWS services. This includes patching the infrastructure software and configuring infrastructure devices. As a customer, you are responsible for implementing best practices for data encryption, patching guest operating system and applications, identity and access management, and network & firewall configurations.
A retail company sells electronic products online and is facing inventory management issues. They want to implement a demand prediction model to address this problem. Which stage in implementing the model is most critical for addressing stockouts and overstock situations?
a. Model training
b. Data collection
c. Business evaluation
d. Feature engineering
Correct Answer: B
The initial stage in developing ML workloads involves identifying the data required for training and evaluating the performance of an ML model. In a cloud environment, a data lake functions as a central repository for storing all types of structured and unstructured data, regardless of scale.
The option that says: Business evaluation is incorrect. While business evaluation is important for assessing the overall impact of the model on business goals, it is not the most critical stage for directly addressing stockouts and overstock situations. This stage typically comes after the model has been developed and deployed, focusing on evaluating its performance and business impact.
An ML specialist has created a foundation model for a customer support chatbot. The bot’s responses are accurate but don’t align with the company’s formatting. The specialist wants to adjust the responses to match the company’s branding without high costs. Which method will fulfill these requirements in the most cost-efficient manner?
a. Data preprocessing
b. Prompt engineering
c. Feature engineering
d. Hyperparameter tuning
Correct Answer: B
Prompt engineering is a technique used to optimize the input prompts given to a large language model (LLM) or foundation model (FM) in order to achieve desired outputs. This method is particularly effective for guiding the model’s behavior, tone, and style without needing extensive and costly adjustments, such as fine-tuning the model on new datasets. By carefully crafting prompts with appropriate context, keywords, and instructions, users can control how the model responds, ensuring that the output aligns with specific business needs, such as a company’s branding and conversational style.
Which of the following factors should be considered when determining the region in which AWS Resources will be deployed? (Choose TWO)
a. Geographic proximity to the company's location
b. The AWS Region’s security level
c. Data sovereignty
d. The planned number of VPCs
e. Cost
Correct Answer: C, E
Per AWS Best Practices, proximity to your end users, regulatory compliance, data residency constraints, and cost are all factors you have to consider when choosing the most suitable AWS Region.
Which statement is true regarding the AWS Shared Responsibility Model?
a. Responsibilities vary depending on the services used
b. Security of the managed services is the responsibility of the customer
c. Patching the guest OS is always the responsibility of AWS
d. Security of the IaaS services is the responsibility of AWS
Correct Answer: A
Customers should be aware that their responsibilities may vary depending on the AWS services chosen. For example, when using Amazon EC2, you are responsible for applying operating system and application security patches regularly. However, such patches are applied automatically when using Amazon RDS.
An e-commerce company is developing a model using Amazon SageMaker to forecast the probability of a product being returned after purchase. The company owns a labeled dataset containing product categories, prices, customer reviews, and return status stored in an Amazon S3 bucket. What machine learning approach is most appropriate for this task?
a. Transfer learning
b. Few-shot learning
c. Supervised learning
d. Unsupervised learning
Correct Answer: C
Supervised Learning is a machine learning approach where a model is trained on a labeled dataset. This means the dataset contains input-output pairs where the desired output is known. The key advantage of Supervised Learning is that it allows the model to learn the relationship between input features like product categories and prices and output label return status, which enables accurate forecasting of future events.